News Snapdragon X Elite shows 2X higher multi-threaded performance than Apple M2 in new benchmarks — Apple M2 is faster in single-threaded performance,...

Man, can people not wait a few weeks? We’re so close. A virtual machine puts too much burden on a system for performance benchmarks.

Plenty of validated cross-platform native ARM64 benchmarks between macOS and Windows.

PugetBench, Cinebench, Geekbench, Speedometer, etc.

Why not run those?

//

we haven’t seen any real-world tests comparing it to Apple-silicon MacBooks until now

I can’t say I agree that now, suddenly, we have seen real-world tests.

There are few real-world applications that are 1) cross-platform and 2) include native benchmarks that are validated cross-platform.

CPU-Z is not one of them.

Please; just wait a few weeks. Devices are already on the way, if not already, with reviewers.

Then the winner can take a victory lap after reviewers have run their tests.
 
While it's nice that Macs have been fast and power efficient, the main advantage that Macs have is that Macs run macOS instead of Windows.

I'm not sure how making a faster machine that will have to run Windows is going to fix that problem.
 
While it's nice that Macs have been fast and power efficient, the main advantage that Macs have is that Macs run macOS instead of Windows.

I'm not sure how making a faster machine that will have to run Windows is going to fix that problem.
... the main disavantage that Macs have is that Macs run macOS instead of Windows.

In context, thousands of games, software's, including government taxes software in hundreds of countries are Windows exclusives. Macs will serve well for the most basic tasks or very specific ones. For the majority of cases, Windows is necessary.
 
... the main disavantage that Macs have is that Macs run macOS instead of Windows.

In context, thousands of games, software's, including government taxes software in hundreds of countries are Windows exclusives. Macs will serve well for the most basic tasks or very specific ones. For the majority of cases, Windows is necessary.
I gave your comment a "like" because I can appreciate that point of view, because I tend to agree. Although, I'm becoming less confident in this belief.

To test this, what are the workloads that satisfy the "majority of use cases"?

I asked CheatGPT and here are the TOP-LEVEL workloads:

Sure, here are the major use cases or workloads for computing:

1. Data Processing and Analytics
2. Scientific and Technical Computing
3. Enterprise Applications
4. Web and Mobile Applications
5. Networking and Communication
6. Cloud Computing
7. Internet of Things (IoT)
8. Cybersecurity
9. Entertainment and Media
10. Financial Services
11. Healthcare
12. Education and E-Learning


Let's approach each one with a "largest OS base" approach to determine necessity:

Certainly, here is the list with the best supporting operating systems for each workload:

1. Data Processing and Analytics
**OS:** Linux

2. Scientific and Technical Computing
**OS:** Linux

3. Enterprise Applications
**OS:** Windows

### 4. Web and Mobile Applications
**OS:** Linux

5. Networking and Communication
**OS:** Linux

6. Cloud Computing
**OS:** Linux

### 7. Internet of Things (IoT)
**OS:** Linux

8. Cybersecurity
**OS:** Linux

9. Entertainment and Media
**OS:** Linux

### 10. Financial Services
**OS:** Linux

11. Healthcare
**OS:** Windows

12. Education and E-Learning
**OS:** Windows


So, as you can see, the "majority of use cases" argument has fallen apart as of late.

Why is this? The reason is OSS, or Open-Source Software and the collaborative development across academia, to name the leading cause.

Thoughts anyone?
 
  • Like
Reactions: Murissokah
Frankly, this is a garbage article. Benchmarking a native OS against an OS in a virtual machine and comparing the two in this context is ludicrous. And yet the Apple hardware beat the Snapdragon in some benchmarks.

Let's see some apples-to-apples (unintended pun) benchmarks here to compare.

What was the TDP of the tested snapdragon vs the lowest end vs a 1-1.5x generation out of date base model M2? 4x? And it still lost single threaded?

Again, let's see some power-to-power, native benchmarks here before trying to declare an outright win.

~~

I've said it a few times, we don't actually need apple-beating hardware, we just need competative hardware. Right now (pre copilot AI hardware) the windows world really doesn't have a comperable mobile option. To get similar performance you get 1/3-1/2 the battery or to get 2/3 the battery the machine is very slow.

Most mac sales are base macbook airs with base M series chips, not macbook pros with the highest end chips. Snapdragon needs to be within about 80% overall of a macbook air I think. People who prefer windows wont be tempted by mac hardware if they can cross that threshold in performance and battery life. On paper this looks reasonably likely if microsoft's prism works well. Apple hit it out of the park with rosetta2, making it an all-but-seamless transition. The x86-64 on arm64 pre-copilot windows runs is trash. If this isn't improved then these machines will struggle to sell in business applications since business programs tended to lag way way behind for updates like re-building on a new architecture.
 
  • Like
Reactions: Murissokah
I believe they should just wait for the M4 powered MacBook Air to be released before publishing an article on benchmark comparison. This article somewhat unintentionally insinuates that the M2 chip is the latest in Apple chip tech when in reality, it’s almost two years old. On top of that, these benchmarks uses a VM that penalizes on latency and performance making the point of this article moot.
 
Qualcomm are missing the mark by comparing against Apple Silicon.

It is the poor sods who are required to use Windows, who were about to choose between Intel and AMD that they have to win over.

Mac users who are already on Mac will continue to use Mac.
People who are considering leaving Wintel for Apple are doing so because Windows 11 with each new update is getting more alike to the contents of my ostomy pouch than a usable system, not because Apple has good hardware.

Qualcomm has promised Linux support, and for Apple Silicon Macs there is Asahi Linux.
That would be more of an apples-to-apples comparison.
Asahi Linux still does not support the M3 series, but the M2 is fully supported.
For a fair comparison, the Mac should have a 12-core "Pro" processor. (M2 Pro is 8P+4E, M3 Pro is 6P+6E but still faster)
Measure battery life, noise level and lap temperature!
 
Last edited:
Frankly, this is a garbage article. Benchmarking a native OS against an OS in a virtual machine and comparing the two in this context is ludicrous. And yet the Apple hardware beat the Snapdragon in some benchmarks.
I would've chosen milder wording, but fully agree with the sentiment. If you benchmark native CPU performance against VM CPU performance you create irrelevant data. Which should be obvious to a PC hardware media company.
 
So, as you can see, the "majority of use cases" argument has fallen apart as of late.

Why is this? The reason is OSS, or Open-Source Software and the collaborative development across academia, to name the leading cause.

Thoughts anyone?
I agree with the observation, but I think cloud services and containerization play a bigger role than OSS and academia.

Cloud services made it so most productivity software can be used by anyone with a browser. Outlook is still not supported in my Ubuntu notebook, but now it doesn't matter. Our finance and accounting software, CRM and ITSM are all SaaS. This is huge for Linux and Mac.

Then containerization made OSS adoption much broader on the service provider side. Why would I run a 50~80 GB licensed Windows Server machine, which I have to mantain and update myself, if all I want is to serve a simple web application? It used to be commonplace that we would find a setup like that with IIS and an oversized server runing a 10mb application. Now you can do that in a 200mb nginx image that's maintained by them, with no licensing. All you need is some basic understanding of docker. Then add the operational versatiliy of container orchestrators, like Kubernetes, and it's very hard to beat.

Considering Microsoft now bundles their own Linux kernel in Windows, choosing an OS for work has become more about preference than anything else, at least for me.
 
  • Like
Reactions: brandonjclark
Just wait for the OEMs to release this new great chip in their old and tired creaky plastic boxes, with inadequate cooling, dodgy keyboards, wobbly trackpads with low precision and huge latency, and with a pair of fans whining at 40dB from startup to shutdown. Mac is not great because M2 is fast. It's great because it's noiseless, without latency, with great keyboard, screen and battery life.
 
We've just about reached maximum lunacy in this article.

🤡 Hoist the Big Top because the circus is in town! 🤡
tom'sHARDWARE is using a post on X where a user runs a "benchmark" on a 2 year old Mac, running a hypervisor, so it can run Windows 11 (ARM) to run a POC benchmarking app against an SOC that isn't even on store shelves yet, that has 50% more cores, all of which are performance cores, where the SOC can boost (for shot periods of time) to use 400% more power, has active cooling and isn't forced to run a second entire OS on a hypervisor that can't even leverage all of the host's (the M2's) cores and memory because it requires dedicated resources for itself.

Apple M2
ADVANTAGE​
Snapdragon X Elite X1E-78-100​
4 x Performance Cores
4 x Efficiency Cores
X1E-78-100 has 3x # of performance cores 👉​
👈M2 has 4x # of efficiency cores :crazy:
12 x Performance Cores
0 x Efficiency Cores​
20W TDP (PL1)
X1E-78-100 has +3W of base TDP 👉​
23W TDP (PL1)​
0W TDP (PL2)
X1E-78-100 has +80W of "burst" TDP2 👉​
80W TDP (PL2)​
macOS 14.5 base OS
🔥who cares🔥​
Windows 11 (ARM) base OS​
Parallels Desktop 19 hypervisor
X1E-78-100 isn't forced to run hypervisor 👉​
no hypervisor​
Windows 11 (ARM) running in hypervisor
X1E-78-100 isn't forced to run second OS 👉​
no OS running in hypervisor​
5nm process node
X1E-78-100 uses more advanced process node 👉​
4nm process node​
2 year old architecture
X1E-78-100 still doesn't ACTUALLY exist for consumers 👉​
2+ weeks until it is on store shelves​
 
Just wait for the OEMs to release this new great chip in their old and tired creaky plastic boxes, with inadequate cooling, dodgy keyboards, wobbly trackpads with low precision and huge latency, and with a pair of fans whining at 40dB from startup to shutdown. Mac is not great because M2 is fast. It's great because it's noiseless, without latency, with great keyboard, screen and battery life.
Do you live in 2014 or something? Windows laptops have been competitive and even beating macs in some of those categories when you compare modern machines. Mac aren't even noise less when you push em to higher wattage. As for screen, battery, and keyboard, notebookcheck and other sources show PLENTY of reviews in the last 3-4 years of windows laptops besting the latest macbook at that time.
keyboard: https://www.tomsguide.com/opinion/i-review-laptops-for-a-living-and-these-have-my-favorite-keyboards
screen: https://www.notebookcheck.net/LG-gram-Style-16Z90RS.722490.0.html
battery: =343149&specs[]=344585]https://www.notebookcheck.net/index.php?id=127065&specs[]=343149&specs[]=344585
 
... the main disavantage that Macs have is that Macs run macOS instead of Windows.

In context, thousands of games, software's, including government taxes software in hundreds of countries are Windows exclusives. Macs will serve well for the most basic tasks or very specific ones. For the majority of cases, Windows is necessary.
Strangely enough, Mac runs Windows faster than Intel chips do. The reason that almost no one bothers to do that, though, is that the Windows OS has very few advantages (mainly, games), and lots of disadvantages. (I mainly have Windows on Mac for games.)

I'm typing this on a Mac, but I was using Windows for most of the evening (POE on a 7800x3d). I have 4 Windows PCs and 2 Macs, and a bunch of Linux servers. For getting work done, Windows is about the worst environment there is, and I've been building Windows (and other) software for 30+ years now and programming for longer than that, so I guess I have some experience on the topic.